gregoa's blog

I've upgraded the kernel on both my laptop (on tuesday) & my laptop (today) to 2.6.26; no problems so far, the extra modules built & the kernels run. — ath5k is still not working with my wifi chip (AR5212, cf. #463425, although the errors look slightly different [0]; let's wait until 2.6.26-686 hits unstable).

on other news: cpu frequency scaling seems to finally work on my laptop (thinkpad R60e with a Intel(R) Celeron(R) M CPU 420 @ 1.60GHz cpu); the trick was to set the minimal frequency explicitly to 400 khz (200 would be the lowest value).

# cat /etc/modules
..
p4-clockmod
cpufreq_ondemand



# cat /etc/sysfs.conf
..
devices/system/cpu/cpu0/cpufreq/scaling_governor = ondemand
devices/system/cpu/cpu0/cpufreq/scaling_min_freq = 399999

updates:

• I have strange network problems here at DebCamp with 2.6.26 & madwifi, fortunately I still have some other kernels installed :)
• ath5k still doesn't work with linux-image-2.6.26-1-686 from unstable.

update:
w00t! ath5k works with linux-image-2.6.26-1-686 2.6.26-2!
but I still have massive packet loss wih both madwifi and ath5k here at DebConf. — & also here some days later at "Hostal Sebas" in BsAs.

[0]

# grep ath5k /var/log/syslog.0
Jul 30 00:36:09 nerys kernel: [  320.450984] ath5k_pci 0000:03:00.0: registered as 'phy0'
Jul 30 00:36:09 nerys kernel: [  321.121107] ath5k phy0: Atheros AR5424 chip found (MAC: 0xa3, PHY: 0x61)
Jul 30 00:36:22 nerys kernel: [  333.513763] ath5k phy0: noise floor calibration timeout (2412MHz)
Jul 30 00:36:22 nerys kernel: [  333.513794] ath5k phy0: can't reset hardware (-11)
Jul 30 00:36:38 nerys kernel: [  349.655496] ath5k phy0: noise floor calibration timeout (2412MHz)
Jul 30 00:36:38 nerys kernel: [  349.655525] ath5k phy0: ath5k_chan_set: unable to reset channel (2412 Mhz)
Jul 30 00:36:42 nerys kernel: [  353.004057] ath5k phy0: noise floor calibration timeout (2412MHz)
Jul 30 00:36:44 nerys kernel: [  355.314594] ath5k phy0: noise floor calibration timeout (2417MHz)

finally I managed to book my flight to DebConf8.

2008-08-03 IB3537
07:45 / Munich, Franz Josef Strauss, Terminal 1
10:25 / Madrid, Barajas, Terminal 4



2008-08-03 IB6845
12:25 / Madrid, Barajas, Terminal 4S
19:40 / Buenos Aires, Pistarini, Terminal A



2008-08-25 IB6844
21:35 / Buenos Aires, Pistarini, Terminal A
14:30 day +1 / Madrid, Barajas, Terminal 4S



2008-08-26 IB3564
16:30 / Madrid, Barajas, Terminal 4
18:55 / Munich, Franz Josef Strauss, Terminal 1

& I'm still surprised how frequently the prices for the same flights change ...

• since last weekend both my desktop & my laptop run linux 2.6.25 & perl 5.10; no remarkable changes/problems so far.
• more interesting: by chance (well, by running powertop) I discovered that my laptop (Thinkpad R60e, Intel Celeron M 420 CPU) now supports cpu frequency scaling; it uses the p4-modclock kernel module & works fine with the ondemand governor. — so there might have been some changes in the kernel which I had missed for a whole week :)

for those interested in cpu frequency scaling I can still recommend john goerzen's fine tutorial.

update 1:
or not. the machine freezes every other minute for several minutes, both with the ondemand & the conservative governor; & also with the userspace governor.

update 2:
after disabling some cpu power management option in the BIOS cpu frequency scaling seems to work, although I see occasional short freezings.

update 3:
well, having freezes of 2-3 seconds each minute is better than everything before but not good enough for actually using it.

when I woke up today (after sleeping in for the first time with my new roll-top in front of my bedroom window) I was surprised & confused by a couple of "congratulations!" messages in my irc away-log. it took me a bit of time & coffee (& looking into my mailbox) to begin to realize that my Debian account had indeed been created while I was asleep. — in fact I guess I still haven't completely realized my new status as Debian Developer.

as others I'd like to follow the good tradition of taking the opportunity to thank some of the people who helped me to get there:

• first & foremost Tony Mancill, my advocate, permanent sponsor, & long-time co-maintainer; for all his support in preparing & quickly uploading packages, tracking down bugs, encouraging & advocating me but also for our very pleasant "chats" via email. — we have already agreed to continue co-maintaining each other's packages.
• the friendly & helpful guys from the Debian Perl Group: Gunnar Wolf, for always encouraging me (to join the group, to apply for the New Maintainer Process, ...); Damyan Ivanov, for teaching me so much (even if he called it "nitpicking"); Gunnar, Damyan, Krzysztof Krzyżaniak, Niko Tyni, Roberto C. Sánchez, Stephen Gran, Frank Lichtenheld, Jaldhar H. Vyas, Russ Allbery, Raphaël Hertzog, & some others for uploading packages I've prepared & helping me when I had questions or problems; Martín Ferrari, for all his help, expertise, & especially for the fun & the good times we share on IRC.
• my Application Manager Wouter Verhelst; for guiding me through the New Maintainer process not only fast but also in a very helpful, at the same time demanding & supporting way.
• the Debian Women sub-project; for making Debian a better & friendlier place not only for women but in general.
• finally: the people who used the time while I was asleep to actually create my account tonight :)

some final thoughts about the NM process from my point of view:

• it took me some time to actually apply for NM; what deterred me was not that I knew it would take some time but that I didn't know how long the time would be; & that I knew that the bigger part of the overall time would consist of waiting.
• I applied on 2007-04-03, my account was created on 2008-04-18; 380 days is not bad altogether in my opinion; the actual work with my AM was from 2007-08-12 until 2007-11-28 (i.e. 108 days, or 28% of the whole duration).
• I did enjoy the actual checks; I had to read & think a lot, & I learned a lot in that period.
• Front Desk (i.e. Christoph Berg) was very quick on all necessary steps.
• I was never demotivated about my Debian work because of all the people supporting me & taking the burden of uploading packages I had prepared. but I have to admit that I got a little bored in the last months of waiting after the report had been submitted by my Application Manager.

& now it's time first to celebrate & then to try to fully grasp my new rights & responsibilities.

first I was wondering why viewing photos with qiv takes so long. then I was wondering about the high incoming network traffic; netstat showed that it was coming from my laptop to my desktop machine.

then I realized that I had started qiv in the xterm that contained the ssh connection to the screen session on my laptop.

on 2007-04-03 I applied to become a Debian Developer; that means that today's my first anniversary of being in the New Maintainer process.

(details on the past & future steps & stages of my NM process can be found on my NM page).

update: DAM have reviewed my application on 2008-04-07, so "only" the actual account creation is missing.

after my laptop I tried to get 2.6.24 running on my desktop machine; several attempts (Debian kernels, self-compiled ones with different options) all led to the same results: a machine that feels rather unresponsive & where music has sporadic dropouts.

some searching on the intarweb showed that the new CFS scheduler (cf. Documentation/sched-design-CFS.txt) (at least the enabled-by-default CONFIG_FAIR_USER_SCHED flavour) might have latency problems, in itself or in combination with the ondemand CPU frequency scaling governor or with nice'd processes; after turning off all CONFIG_FAIR_* options in .config, recompiling & rebooting the machine seems to behave normally.

maybe I'm missing some great new features but actually I prefer a working desktop computer :)

update 1: approximately half an hour after the original blog entry I had my first "ion sorbet", i.e. a half-frozen ion3 (the applications are still running, I can switch to a tty but I cannot change/close/whatever ion3's frames). at the moment I have no idea what's going (wr)on(g) here ...

update 2: two more "ion sorbets" today yet; let's see if a new kernel (CONFIG_FAIR_USER_SCHED seems to be fixed in 2.6.24.1) & a new ion3 version help.

update 3: still the same problems (& inbetween some glitches with switching from X to a tty and back; I hope I solved them with some kernel config options); ATM I suspect ion3 to be the culprit; when it hangs it eats almost all the CPU, & strace'ing the pid gives only SIGALRM messages.

update 4: maybe I fixed the problem; by changing from ion3's deprecated statusbar_external.lua to the newer statusd_exec.lua (I just wanted to get rid of the error when starting ion3 ...).

update 5: & after compiling a kernel without PREEMPT ion3-statusd doesn't get stuck at using all my cpu every once in a while. or after removing some duplicated statusd_*.lua scripts. who knows.

update 6: nope, ion3-statusd is still hogging the cpu. argl.

update 7: I guess it's statusbar_external.lua's fault, as it happens on my other machine too.

no big step for humankind but I'm quite happy that linux 2.6.24 is finally running on my laptop (thinkpad R60e). the culprit that caused the freezes during earlier attempts was cpudyn; after deactivating it I had no more problems so far.

what's not working is ath5k which seems to dislike my wifi card; but the recent madwifi packages do what they are supposed to do.

JFTR:

• I grabbed the latest kernel snapshot from the Debian kernel team's repo.
• after reading #463353 I downloaded kel's source packages for madwifi-*, built them, installed them, & built the modules as usual with m-a.

now let's have a look at 2.6.24 on my desktop machine :)

finally I found the time to write down the configuration we use on our server to blacklist hosts that try to send emails to pre-defined nonexistent mail addresses. it uses only exim & postgresql & is (except for adding the "triggering" recipient email address patterns to the respective table) rather self-contained.

the documentation can be found in the article dynamic blacklisting with exim & postgresql. feel free to use/improve/comment/... the ideas!

an hour ago I received an email from my AM who considers my packages to be in good shape; that means that the "Tasks and Skills Check" part of my NM process is finished. & wouter has also already sent the NM report mail including his recommendation to accept me as a DD to debian-newmaint! so only FD review, DAM review, & account creation are lying ahead of me.

I'm indeed quite happy tonight :)